Revealing design archetypes and flexibility in e-molecule import pathways using Modeling to Generate Alternatives and interpretable machine learning

(en) Given the central role of green e-molecule imports in the European energy transition, many studies optimize import pathways and identify a single cost-optimal solution. However, cost optimal designs may be sensitive to real-world constraints, as regulatory, spatial, and stakeholder considerations are often difficult to represent in optimization models and may affect their practical feasibility. To address this limitation, we use a greenfield, snapshot capacity expansion model of a hydrogen production and transport supply chain from Morocco to Belgium, considering hydrogen, ammonia, methane, and methanol as import carriers under a fixed hydrogen demand. Using the Modeling All Alternatives (MAA) method, which is a Modeling to Generate Alternatives (MGA) approach, we generate a large and diverse set of near-optimal solutions within an acceptable cost margin, enabling the exploration of design flexibility under structural uncertainties. Given the complexity of interpreting a large number of near-optimal alternatives, interpretable machine learning is applied to extract actionable insights from the large solution space. Results show that hydrogen transported via pipelines is the lowest-cost option (91 €/MWh), while methane transported via pipelines is the most expensive (213 €/MWh). Moreover, MGA reveals a broad near-optimal space with great flexibility: solar, wind, and storage are not strictly required to remain within 10% of the cost optimum. Wind capacity can reach up to 20 GW, PV capacity up to 40 GW, battery capacity up to 60 GWh, and hydrogen storage up to 162 GWh, depending on the carrier. Limited access to wind energy favors systems with higher electrolyzer and solar capacities combined with storage, particularly in methanol pathways. On the other hand, limited storage availability favors wind-dominated systems, with ammonia or methane pathways becoming more attractive. These results demonstrate that multiple structurally distinct supply chains are economically viable, giving stakeholders broad room for decision-making.

Affiliations

UCLouvainSST/IMMC/TFL - Thermodynamics and fluid mechanics

Citations

APA
Chicago
FWB

Kchaou, M. (2026). Revealing design archetypes and flexibility in e-molecule import pathways using Modeling to Generate Alternatives and interpretable machine learning. Power-to-X, 100003. https://doi.org/10.1016/j.ptx.2026.100003 (Original work published 2026)